World Engines Lab
with Delta_Ark Studios
Impact
As autonomous agents become more prevalent in our current media, urban, industrial and military landscapes, citizens will need to develop a much deeper understanding of how to engage with them, think with/of them and criticize them.
Citizens will also have to develop new ideas about responsibility, autonomy and identity as these machines automate cognitive processes that formerly were only embodied in human (and animal) bodies.
Consequently, The World Engines Lab is building a cutting edge simulation along with concomitant artworks born directly from within it in order to develop citizens’ agentic literacy by exploring the role of autonomous machines in everyday life in the near, medium and far future.
Overview
The Lab, led by Delta_Ark Studios and incubated at Gray Area, is focused on making a large persistent simulation with powerful agents that artists can use/collaborate with for live performance.
This world can be adapted to create games, films, sculptures, agent-based scientific and interaction experiments and artificial life.
Researcher-artists in the lab will use these systems to explore multi-species relations (humans, AI) and the future of evolution (on this planet, or others), articulating different narratives and ramifications of human-agent interaction.
Techniques
The studio is currently working to create:
1) A dedicated server that can host a game world persistently. Ideally, this world will feature dynamic flora and fauna populations with genetic change.
2) Inside this world will be a suite of agents, built off a modifiable template agent that connects Unreal’s native AI systems with GPT (for extemporaneous behavior and dialogue).
3) We are also exploring the latest AI techniques for live-motion capture, so that the artists in the lab can do live performances in world or, as a stretch goal, in hybrid formats.
People
The lab is currently run by Ari Kalinowski (Delta_Ark). The agent technical advisor is Batu Aytemiz. The motion capture technical advisor is Daniel Sierra.
Parish Bracha and Alexa Ann Bonomo are the current motion researchers. Arden Schager and Kat Zhang are the current agent researchers. Will Freundenheiim and Yvonne Fang are the current lab observers.
Alasdaire Milne (Serpentine) and Hannah Scott (Gray Area) are the current strategic advisors. We are currently looking for a humanoid character designer advisor and researcher and a monster/machine character designer and researcher.
Outreach
Every six months, the lab will open to the public in order to (a.) show works in an exhibition format and (b.) other works as public performances and (c.) the current state of our simulated world platform.
Additionally, (d.) we will host a monthly workshop (“Simulation Sessions”) at Gray Area, allowing local artists to test the platform and (e.) publish a research report on the current state of the lab and research.
In the best case scenario, we will also release parts of the world as an Unreal Engine plugin, so others can build on top of it.
Simulation Sessions
Simulation Sessions is a free monthly event series facilitated by the World Engines Lab, incubated at Gray Area. In response to the growing prevalence and sophistication of virtual simulations and environments, these sessions seek to sustain a local community of 3D designers, artists, and technologists, facilitate peer teaching and learning, and spark dialogue about the social, cultural, and political dimensions of building digital worlds.
All, regardless of skill or experience level, are welcome to attend and bring their current projects (in any 3D platform, including Unreal, Unity, Godot, Three.js, Babylon.js, Rhino, Blender, Houdini, Spline, Womp, TouchDesigner, Custom 3DAI tools, and more).
Each session begins promptly at 6pm with a brief talk or mini-workshop led by a member of the community. At 6:30pm, participants are invited to work on projects, seek technical advice, mingle, and meet others engaged in the practice of building simulations.
To RSVP for the next Simulation Session, visit our events page, or subscribe to our newsletter to stay informed about upcoming sessions.
Output
Contra Autonomous Warfare
A music video for a longer piece entitled Novascene Hekhalot made entirely by filming agents.
Binah
An open world game that is built and re-built by agents during gameplay, opening and closing different areas on different playthroughs.
GameAI Research: Real-time Construction Robotics
A real-time robotic, agent-based terraforming system
ML Research: Agent Needs and LLM Dialogue System
An implementation of in-world LLM dialogue system for agents that navigate the space, trying to fulfill their basic needs
Lead Investigator
Delta_Ark (Ari Kalinowski)
Delta_Ark’s current work is a sequence of interactive virtual environments and audio-visual experiences that represent and embody different relationships between human-like entities, artificial intelligences (of various powers) and natural systems (or chimeras) in different science fiction contexts, often involving precipitous technological evolution, climate change or the far future. Many of these works explore future power-relations between AGIs, ASIs and cyberized forms of collective and individual human intelligence in order to sketch out some visions of a post-biological (and/or synthetic-biological) future. Motifs from Jewish mythology, Japanese popular media and contemporary philosophy often form the contextual backdrop of these explorations.
Advisors
Alasdair Milne
Alasdair’s work focuses on the collaborative systems that artists build around new technologies. As part of Serpentine’s Creative AI Lab he studies how machine learning is increasingly integrated into artistic workflows, becoming part of broader and more complex systems of tooling. He researches curatorial methods for how best to support such emerging practices, records histories of their construction (such as with 0rphan Drift & Etic Lab’s ISCRI, and dmstfctn’s God Mode) and develops practice-informed theories in parallel.
Alasdair has focused on developing collaborative spaces and methods for in-depth conceptual exploration, a research process which feeds forward into the technical development of artistic projects. As established support structures in AI art practice increasingly focus on the technical, this curatorial method for ‘conceptual R&D’ aids artists to bridge the complexities of machine learning as it is deployed in specific cultural contexts.
More broadly, Alasdair’s focus on transdisciplinary collaboration makes Gray Area a vital case study towards an understanding of institutional infrastructures for ambitious artistic projects that draw together expertise not just between the artistic and the technical, but also from diverse disciplines.
If you are an artist working with machine learning in the Bay Area, feel free to reach out to Alasdair directly at [email protected].
